Valkey

Valkey is an in-memory key-value server whose code descends from Redis and still carries the 3-Clause BSD licence. What sets it apart is who stands behind it: the Linux Foundation backs the project, LF Projects, LLC holds the trademarks, and companies such as AWS, Google Cloud, Alibaba Cloud, Oracle, Percona, ByteDance and Ericsson take part.

The software is not sold by the project. You either run it yourself as a standalone daemon or a cluster, or rent it from a hosting provider that sets its own prices. The homepage names no country of establishment for the project or its governing body. For a buyer, the real decision is therefore which provider or hosting arrangement to pair with a neutral, community-governed codebase.

• Familiar data types. Values can be strings, numbers, hashes, lists, sets, sorted sets, bitmaps or HyperLogLogs. That range lets one server hold counters, queues and ranked lists side by side, without bolting a second store onto the application.
• Module plugins. Modules can add new commands and new data types, so the server can be extended without forking it. The project's homepage presents this as the route for specialised needs that the built-in types do not cover. Third parties can therefore ship commands without waiting for the core project.
• Per-database access control. Access control lists apply per database, which allows different applications to share one server with separate permissions. The pages read give no further detail on how rules are written or audited. Rules apply to each database rather than the whole server.
• Lua as an optional module. From version 9.1, Lua scripting is a module and can be removed when nobody uses it. Dropping an unused component shrinks what a security reviewer has to consider, which is the stated attraction.
• Memory savings in 9.1. The project states that version 9.1 cuts memory use by up to 20% for small strings. That is the project's own figure, so a team holding many short keys should measure it on its own data.
• Clustering and replication. Replication and high-availability options exist for running Valkey as a cluster instead of a single daemon. The homepage lists them but the pages we read do not explain failover timing or topology limits.

• A platform team avoiding single-vendor control. Foundation governance and a BSD licence suit organisations that want the key-value layer to outlive any one company's licensing decisions. The participation list on the homepage shows several large cloud providers contributing rather than one owner.
• A Kubernetes shop running its own infrastructure. The homepage mentions Solanica, which lets users run Valkey on their own Kubernetes clusters. A team that already operates clusters can keep session or cache data inside its own environment rather than renting a hosted service.
• A company already buying cloud services. Amazon ElastiCache, Google Cloud Memorystore, Aiven, DigitalOcean, Heroku Key-Value Store, Oracle OCI Cache and IONOS Cloud are named as hosts. A buyer can compare those providers on price and region while the data model stays the same.

The pages we read give no installation steps, so this is what they do state. Valkey can be self-hosted as a standalone daemon or as a cluster with replication, or taken as a managed service from one of the named providers.

Self-hosting needs your own server or Kubernetes cluster and someone to operate it. Hosted routes need an account with the chosen provider and agreement to that provider's pricing. The project names no free trial because the software itself is open source under BSD.

Valkey carries BSD-licensed code inherited from Redis and sits under the Linux Foundation, with AWS, Google Cloud, Alibaba Cloud, Oracle and others taking part, so no single company controls it.

You run it as a standalone daemon or cluster, or rent it from providers such as ElastiCache, Aiven or IONOS Cloud. The project itself offers no hosted service, support contract or published country of establishment.

The database management systems list places Valkey 21 of 22: it follows DuckDB and comes before YugabyteDB. After the challenger places, Valkey is one of the established names, ranked in order and without a score.

The vendor states no country of establishment, so Valkey is left out of the regional count for this list.

Data types cover strings, numbers, hashes, lists, sets, sorted sets, bitmaps and HyperLogLogs, and modules can add commands and types.

Version 9.1 moves Lua scripting into a module that can be dropped, and the project says it cuts memory for small strings by up to 20%.

Replication and high-availability options exist for clusters.

Pricing as recorded: Open-source software with no price list; managed hosting sold separately by cloud providers.

The project lists no pricing plans, because Valkey is open-source software rather than a product sold by its maintainers. The pages we read name no trial and no price for the code. Costs arise elsewhere: server or Kubernetes capacity if you host it, or a bill from a provider if you rent it. Amazon ElastiCache, Google Cloud Memorystore, Aiven, DigitalOcean, Heroku Key-Value Store, Oracle OCI Cache and IONOS Cloud are named as sellers of managed hosting, each with its own pricing that we have not quoted here. Support is likewise a matter between you and your host, as the project itself describes no support contract.

Checked against valkey.io for Best Database Management Systems in 2026, review date 7 November 2026.

Security detail on the pages read is brief. Access control lists work per database, and server logs are produced in JSON, which is easy to feed into a log pipeline. Lua scripting became a module in version 9.1 and can be left out if unused.

The pages do not describe encryption, certifications, audit reports or data-residency choices. Where the data lives depends on your deployment or your host. The homepage and governance page show no establishment country for the project or its governing body, so jurisdiction questions need to be settled with the provider you choose.

Integrations the vendor names (8):

• Amazon ElastiCache
• Google Cloud Memorystore
• Aiven
• DigitalOcean
• Heroku Key-Value Store
• Oracle OCI Cache
• IONOS Cloud
• Solanica (Kubernetes)

These are the nearest neighbours of Valkey in the same list, picked on shared traits and price model.

• FerretDB (#13 of 22): A good match for: Teams keeping MongoDB drivers who want an Apache-licensed server.
• Google Firestore (#16 of 22): Serverless document database on Google Cloud with an optional MongoDB-compatible API. Best for: Google Cloud teams who want a document store with no capacity planning.
• Amazon DocumentDB (#17 of 22): Best for: AWS teams that want to keep MongoDB drivers under an AWS invoice.
• YugabyteDB (#22 of 22): Managed distributed SQL billed by vCPU, with add-ons for security and recovery. Best for: Teams sizing a managed SQL cluster by vCPU who can estimate storage and transfer.

Read from the vendor’s own pages on 7 November 2026.

• valkey.io: BSD licence, Linux Foundation governance and named participants, data types, per-database ACLs, JSON logs, version 9.1 changes, the 2.1 million requests per second claim, hosting providers and absence of pricing or a stated country

We test for four things rather than for features: setup time, price in practice, exit and audience. On those tests Valkey lands at place 21 of 22. Details: how we rank software.